From 9235609c2e56b960d9a3f6bb3eacfd54f10b6191 Mon Sep 17 00:00:00 2001
From: Lea Silakov <me@ileamare.ru>
Date: Sun, 1 Mar 2026 19:24:03 +0300
Subject: [PATCH 1/5] naive version (94s)

---
 app/Parser.php | 35 ++++++++++++++++++++++++++++++++++-
 1 file changed, 34 insertions(+), 1 deletion(-)

diff --git a/app/Parser.php b/app/Parser.php
index b74cf7b9f7..dc48b5cd69 100644
--- a/app/Parser.php
+++ b/app/Parser.php
@@ -8,6 +8,39 @@ final class Parser
 {
     public function parse(string $inputPath, string $outputPath): void
     {
-        throw new Exception('TODO');
+        $out = [];
+
+        $handle = fopen($inputPath, 'r');
+        if ($handle === false) {
+            throw new Exception("Failed to open input file: {$inputPath}");
+        }
+
+        try {
+            while (($line = fgets($handle)) !== false) {
+                $line = rtrim($line, "\r\n");
+                if ($line !== '') {
+                    [$path, $date] = $this->parseLine($line);
+                    if (!isset($out[$path])) $out[$path] = [];
+                    $out[$path][$date] = ($out[$path][$date] ?? 0) + 1;
+                }
+            }
+        } finally {
+            fclose($handle);
+        }
+
+        foreach ($out as $path => &$dates) {
+            ksort($dates);
+        }
+
+        file_put_contents($outputPath, json_encode($out, JSON_PRETTY_PRINT));
+    }
+
+    private function parseLine(string $line): array
+    {
+        // skipping https://stitcher.io/, , it's all the same domain anyway
+        // also skipping 15ch of datetime
+        $line = substr($line, 19, -15);
+
+        return explode(',', $line);
     }
 }
\ No newline at end of file

From 59492e985c7cd33d1dbb903b06d8839602573e13 Mon Sep 17 00:00:00 2001
From: Lea Silakov <me@ileamare.ru>
Date: Sun, 1 Mar 2026 19:58:34 +0300
Subject: [PATCH 2/5] better date processing and custom output (73s)

---
 app/Parser.php | 135 ++++++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 121 insertions(+), 14 deletions(-)

diff --git a/app/Parser.php b/app/Parser.php
index dc48b5cd69..d2d9751209 100644
--- a/app/Parser.php
+++ b/app/Parser.php
@@ -6,41 +6,148 @@
 
 final class Parser
 {
+    private const int READ_CHUNK = 163_840;
+    private const int WRITE_BUF = 1_048_576;
+
     public function parse(string $inputPath, string $outputPath): void
     {
+        gc_disable();
+
+        $fileSize = filesize($inputPath);
+        
+        $dateIds = [];
+        $dates = [];
+        $dateCount = 0;
+        for ($y = 21; $y <= 26; $y++) {
+            for ($m = 1; $m <= 12; $m++) {
+                $maxD = match ($m) {
+                    2 => $y === 24 ? 29 : 28,
+                    4, 6, 9, 11 => 30,
+                    default => 31,
+                };
+                $mStr = ($m < 10 ? '0' : '') . $m;
+                $ymStr = "{$y}-{$mStr}-";
+                for ($d = 1; $d <= $maxD; $d++) {
+                    $key = $ymStr . (($d < 10 ? '0' : '') . $d);
+                    $dateIds[$key] = $dateCount;
+                    $dates[$dateCount] = $key;
+                    $dateCount++;
+                }
+            }
+        }
+
         $out = [];
 
-        $handle = fopen($inputPath, 'r');
+        $handle = fopen($inputPath, 'rb');
         if ($handle === false) {
             throw new Exception("Failed to open input file: {$inputPath}");
         }
+        
+        stream_set_read_buffer($handle, 0);
 
         try {
-            while (($line = fgets($handle)) !== false) {
-                $line = rtrim($line, "\r\n");
-                if ($line !== '') {
-                    [$path, $date] = $this->parseLine($line);
-                    if (!isset($out[$path])) $out[$path] = [];
-                    $out[$path][$date] = ($out[$path][$date] ?? 0) + 1;
+            $remaining = $fileSize;
+            
+            while ($remaining > 0) {
+                $toRead = $remaining > self::READ_CHUNK ? self::READ_CHUNK : $remaining;
+                $chunk = fread($handle, $toRead);
+                $chunkLen = strlen($chunk);
+                
+                if ($chunkLen === 0) break;
+                $remaining -= $chunkLen;
+                
+                $lastNl = strrpos($chunk, "\n");
+                if ($lastNl === false) {
+                    $this->processChunk($chunk, $dateIds, $out);
+                    continue;
+                }
+                
+                $this->processChunk(substr($chunk, 0, $lastNl + 1), $dateIds, $out);
+                
+                $tail = $chunkLen - $lastNl - 1;
+                if ($tail > 0) {
+                    fseek($handle, -$tail, SEEK_CUR);
+                    $remaining += $tail;
                 }
             }
         } finally {
             fclose($handle);
         }
 
-        foreach ($out as $path => &$dates) {
-            ksort($dates);
+        foreach ($out as $path => &$dateIdCounts) {
+            ksort($dateIdCounts);
         }
+        unset($dateIdCounts);
 
-        file_put_contents($outputPath, json_encode($out, JSON_PRETTY_PRINT));
+        $this->jsonize($outputPath, $out, $dates);
     }
-
-    private function parseLine(string $line): array
+    
+    private function processChunk(string $chunk, array $dateIds, array &$out): void
     {
-        // skipping https://stitcher.io/, , it's all the same domain anyway
+        $pos = 0;
+        $chunkLen = strlen($chunk);
+        
+        while ($pos < $chunkLen) {
+            $nlPos = strpos($chunk, "\n", $pos);
+            if ($nlPos === false) {
+                $line = substr($chunk, $pos);
+                if ($line !== '') {
+                    $this->processLine($line, $dateIds, $out);
+                }
+                break;
+            }
+            
+            $line = substr($chunk, $pos, $nlPos - $pos);
+            if ($line !== '') {
+                $this->processLine($line, $dateIds, $out);
+            }
+            $pos = $nlPos + 1;
+        }
+    }
+    
+    private function processLine(string $line, array $dateIds, array &$out): void
+    {
+        // skipping https://stitcher.io/, it's all the same domain anyway
         // also skipping 15ch of datetime
         $line = substr($line, 19, -15);
+        
+        $commaPos = strpos($line, ',');
+        if ($commaPos === false) return;
+        
+        $path = substr($line, 0, $commaPos);
+        $dateFull = substr($line, $commaPos + 1, 10);
+
+        $date = substr($dateFull, 2);
+        
+        $dateKey = $dateIds[$date] ?? $dateFull;
+        
+        if (!isset($out[$path])) {
+            $out[$path] = [];
+        }
+        $out[$path][$dateKey] = ($out[$path][$dateKey] ?? 0) + 1;
+    }
+
+    private function jsonize($filename, &$out, array $dates) {
+        $file = fopen($filename, 'wb');
+        stream_set_write_buffer($file, self::WRITE_BUF);
+        fwrite($file, '{');
+
+        $isFirst = true;
+        foreach ($out as $k => $ds) {
+            $buf = $isFirst ? "\n    \"" : ",\n    \"";
+            $buf .= str_replace('/', '\\/', $k)."\": {\n";
+            $firstDate = true;
+            foreach ($ds as $dateId => $v) {
+                $dateStr = is_int($dateId) ? '20' . $dates[$dateId] : $dateId;
+                $buf .= ($firstDate ? '' : ",\n") . "        \"$dateStr\": $v";
+                $firstDate = false;
+            }
+            $buf .= "\n    }";
+            fwrite($file, $buf);
+            $isFirst = false;
+        }
 
-        return explode(',', $line);
+        fwrite($file, "\n}");
+        fclose($file);
     }
 }
\ No newline at end of file

From 5f6a4aeba8c3b832522bc242998b4e4521904dac Mon Sep 17 00:00:00 2001
From: Lea Silakov <me@ileamare.ru>
Date: Fri, 13 Mar 2026 03:35:35 +0300
Subject: [PATCH 3/5] cleaned up the file read logic and global namespace for
 funcs (31.45s)

---
 app/Parser.php | 150 +++++++++++++++++--------------------------------
 1 file changed, 52 insertions(+), 98 deletions(-)

diff --git a/app/Parser.php b/app/Parser.php
index d2d9751209..5c2e6a040c 100644
--- a/app/Parser.php
+++ b/app/Parser.php
@@ -2,19 +2,18 @@
 
 namespace App;
 
-use Exception;
-
 final class Parser
 {
-    private const int READ_CHUNK = 163_840;
-    private const int WRITE_BUF = 1_048_576;
+    private const READ_CHUNK = 163_840;
+    private const PREFIX_LEN = 25; // "https://stitcher.io/blog/"
+    private const WRITE_BUF  = 1_048_576;
 
     public function parse(string $inputPath, string $outputPath): void
     {
-        gc_disable();
+        \gc_disable();
+
+        $fileSize = \filesize($inputPath);
 
-        $fileSize = filesize($inputPath);
-        
         $dateIds = [];
         $dates = [];
         $dateCount = 0;
@@ -38,116 +37,71 @@ public function parse(string $inputPath, string $outputPath): void
 
         $out = [];
 
-        $handle = fopen($inputPath, 'rb');
-        if ($handle === false) {
-            throw new Exception("Failed to open input file: {$inputPath}");
-        }
-        
-        stream_set_read_buffer($handle, 0);
-
-        try {
-            $remaining = $fileSize;
-            
-            while ($remaining > 0) {
-                $toRead = $remaining > self::READ_CHUNK ? self::READ_CHUNK : $remaining;
-                $chunk = fread($handle, $toRead);
-                $chunkLen = strlen($chunk);
-                
-                if ($chunkLen === 0) break;
-                $remaining -= $chunkLen;
-                
-                $lastNl = strrpos($chunk, "\n");
-                if ($lastNl === false) {
-                    $this->processChunk($chunk, $dateIds, $out);
-                    continue;
-                }
-                
-                $this->processChunk(substr($chunk, 0, $lastNl + 1), $dateIds, $out);
-                
-                $tail = $chunkLen - $lastNl - 1;
-                if ($tail > 0) {
-                    fseek($handle, -$tail, SEEK_CUR);
-                    $remaining += $tail;
-                }
+        $handle = \fopen($inputPath, 'rb');
+        \stream_set_read_buffer($handle, 0);
+        $remaining = $fileSize;
+
+        while ($remaining > 0) {
+            $toRead = $remaining > self::READ_CHUNK ? self::READ_CHUNK : $remaining;
+            $chunk = \fread($handle, $toRead);
+            $chunkLen = \strlen($chunk);
+
+            if ($chunkLen === 0) break;
+            $remaining -= $chunkLen;
+
+            $lastNl = \strrpos($chunk, "\n");
+            if ($lastNl === false) break;
+
+            $tail = $chunkLen - $lastNl - 1;
+            if ($tail > 0) {
+                \fseek($handle, -$tail, SEEK_CUR);
+                $remaining += $tail;
+            }
+
+            $p = self::PREFIX_LEN;
+            while ($p < $lastNl) {
+                $sep = \strpos($chunk, ',', $p);
+                if ($sep === false || $sep >= $lastNl) break;
+                $slug = \substr($chunk, $p, $sep - $p);
+                $date = \substr($chunk, $sep + 3, 8);
+                $dateKey = $dateIds[$date] ?? ('20' . $date);
+                if (!isset($out[$slug])) $out[$slug] = [];
+                $out[$slug][$dateKey] = ($out[$slug][$dateKey] ?? 0) + 1;
+                $p = $sep + 52;
             }
-        } finally {
-            fclose($handle);
         }
 
-        foreach ($out as $path => &$dateIdCounts) {
-            ksort($dateIdCounts);
+        \fclose($handle);
+
+        foreach ($out as &$dateIdCounts) {
+            \ksort($dateIdCounts);
         }
         unset($dateIdCounts);
 
         $this->jsonize($outputPath, $out, $dates);
     }
-    
-    private function processChunk(string $chunk, array $dateIds, array &$out): void
-    {
-        $pos = 0;
-        $chunkLen = strlen($chunk);
-        
-        while ($pos < $chunkLen) {
-            $nlPos = strpos($chunk, "\n", $pos);
-            if ($nlPos === false) {
-                $line = substr($chunk, $pos);
-                if ($line !== '') {
-                    $this->processLine($line, $dateIds, $out);
-                }
-                break;
-            }
-            
-            $line = substr($chunk, $pos, $nlPos - $pos);
-            if ($line !== '') {
-                $this->processLine($line, $dateIds, $out);
-            }
-            $pos = $nlPos + 1;
-        }
-    }
-    
-    private function processLine(string $line, array $dateIds, array &$out): void
-    {
-        // skipping https://stitcher.io/, it's all the same domain anyway
-        // also skipping 15ch of datetime
-        $line = substr($line, 19, -15);
-        
-        $commaPos = strpos($line, ',');
-        if ($commaPos === false) return;
-        
-        $path = substr($line, 0, $commaPos);
-        $dateFull = substr($line, $commaPos + 1, 10);
-
-        $date = substr($dateFull, 2);
-        
-        $dateKey = $dateIds[$date] ?? $dateFull;
-        
-        if (!isset($out[$path])) {
-            $out[$path] = [];
-        }
-        $out[$path][$dateKey] = ($out[$path][$dateKey] ?? 0) + 1;
-    }
 
     private function jsonize($filename, &$out, array $dates) {
-        $file = fopen($filename, 'wb');
-        stream_set_write_buffer($file, self::WRITE_BUF);
-        fwrite($file, '{');
+        $file = \fopen($filename, 'wb');
+        \stream_set_write_buffer($file, self::WRITE_BUF);
+        \fwrite($file, '{');
 
         $isFirst = true;
         foreach ($out as $k => $ds) {
-            $buf = $isFirst ? "\n    \"" : ",\n    \"";
-            $buf .= str_replace('/', '\\/', $k)."\": {\n";
+            $buf = $isFirst ? "\n    " : ",\n    ";
+            $isFirst = false;
+            $buf .= "\"\\/blog\\/" . \str_replace('/', '\\/', $k) . "\": {\n";
             $firstDate = true;
             foreach ($ds as $dateId => $v) {
-                $dateStr = is_int($dateId) ? '20' . $dates[$dateId] : $dateId;
+                $dateStr = \is_int($dateId) ? '20' . $dates[$dateId] : $dateId;
                 $buf .= ($firstDate ? '' : ",\n") . "        \"$dateStr\": $v";
                 $firstDate = false;
             }
             $buf .= "\n    }";
-            fwrite($file, $buf);
-            $isFirst = false;
+            \fwrite($file, $buf);
         }
 
-        fwrite($file, "\n}");
-        fclose($file);
+        \fwrite($file, "\n}");
+        \fclose($file);
     }
-}
\ No newline at end of file
+}

From 9e3026202b9b4bd45decd2d44525aa3e38eeeddb Mon Sep 17 00:00:00 2001
From: Lea Silakov <me@ileamare.ru>
Date: Fri, 13 Mar 2026 03:45:49 +0300
Subject: [PATCH 4/5] binary packing dates and paths, also jsonize changes
 (15.9s)

---
 app/Parser.php | 80 +++++++++++++++++++++++++++++++++++++-------------
 1 file changed, 59 insertions(+), 21 deletions(-)

diff --git a/app/Parser.php b/app/Parser.php
index 5c2e6a040c..36e79f72fa 100644
--- a/app/Parser.php
+++ b/app/Parser.php
@@ -2,9 +2,12 @@
 
 namespace App;
 
+ini_set('memory_limit', '8192M');
+
 final class Parser
 {
     private const READ_CHUNK = 163_840;
+    private const PATH_SCAN_SIZE = 2_097_152;
     private const PREFIX_LEN = 25; // "https://stitcher.io/blog/"
     private const WRITE_BUF  = 1_048_576;
 
@@ -35,7 +38,38 @@ public function parse(string $inputPath, string $outputPath): void
             }
         }
 
-        $out = [];
+        $dateIdBytes = [];
+        foreach ($dateIds as $date => $id) {
+            $dateIdBytes[$date] = \chr($id & 0xFF) . \chr($id >> 8);
+        }
+
+        $handle = \fopen($inputPath, 'rb');
+        \stream_set_read_buffer($handle, 0);
+        $raw = \fread($handle, \min(self::PATH_SCAN_SIZE, $fileSize));
+        \fclose($handle);
+
+        $pathIds = [];
+        $paths = [];
+        $pathCount = 0;
+        $pos = 0;
+        $lastNl = \strrpos($raw, "\n") ?: 0;
+
+        while ($pos < $lastNl) {
+            $nlPos = \strpos($raw, "\n", $pos + 52);
+            if ($nlPos === false) break;
+
+            $slug = \substr($raw, $pos + self::PREFIX_LEN, $nlPos - $pos - 51);
+            if (!isset($pathIds[$slug])) {
+                $pathIds[$slug] = $pathCount;
+                $paths[$pathCount] = $slug;
+                $pathCount++;
+            }
+
+            $pos = $nlPos + 1;
+        }
+        unset($raw);
+
+        $buckets = \array_fill(0, $pathCount, '');
 
         $handle = \fopen($inputPath, 'rb');
         \stream_set_read_buffer($handle, 0);
@@ -63,41 +97,45 @@ public function parse(string $inputPath, string $outputPath): void
                 $sep = \strpos($chunk, ',', $p);
                 if ($sep === false || $sep >= $lastNl) break;
                 $slug = \substr($chunk, $p, $sep - $p);
-                $date = \substr($chunk, $sep + 3, 8);
-                $dateKey = $dateIds[$date] ?? ('20' . $date);
-                if (!isset($out[$slug])) $out[$slug] = [];
-                $out[$slug][$dateKey] = ($out[$slug][$dateKey] ?? 0) + 1;
+                if (!isset($pathIds[$slug])) {
+                    $pathIds[$slug] = $pathCount;
+                    $paths[$pathCount] = $slug;
+                    $buckets[$pathCount] = '';
+                    $pathCount++;
+                }
+                $buckets[$pathIds[$slug]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
                 $p = $sep + 52;
             }
         }
 
         \fclose($handle);
 
-        foreach ($out as &$dateIdCounts) {
-            \ksort($dateIdCounts);
-        }
-        unset($dateIdCounts);
-
-        $this->jsonize($outputPath, $out, $dates);
+        $this->jsonize($outputPath, $buckets, $paths, $dates, $dateCount);
     }
 
-    private function jsonize($filename, &$out, array $dates) {
+    private function jsonize(string $filename, array $buckets, array $paths, array $dates, int $dateCount): void {
+        $datePrefixes = \array_map(fn($d) => '        "20' . $d . '": ', $dates);
+        $escapedPaths = \array_map(fn($p) => "\"\\/blog\\/" . \str_replace('/', '\\/', $p) . '"', $paths);
+
         $file = \fopen($filename, 'wb');
         \stream_set_write_buffer($file, self::WRITE_BUF);
         \fwrite($file, '{');
 
         $isFirst = true;
-        foreach ($out as $k => $ds) {
+        foreach ($paths as $pid => $path) {
+            if ($buckets[$pid] === '') continue;
+
+            $hits = \array_count_values(\unpack('v*', $buckets[$pid]));
+            $dateEntries = [];
+            for ($d = 0; $d < $dateCount; $d++) {
+                if (isset($hits[$d]))
+                    $dateEntries[] = $datePrefixes[$d] . $hits[$d];
+            }
+            if (!$dateEntries) continue;
+
             $buf = $isFirst ? "\n    " : ",\n    ";
             $isFirst = false;
-            $buf .= "\"\\/blog\\/" . \str_replace('/', '\\/', $k) . "\": {\n";
-            $firstDate = true;
-            foreach ($ds as $dateId => $v) {
-                $dateStr = \is_int($dateId) ? '20' . $dates[$dateId] : $dateId;
-                $buf .= ($firstDate ? '' : ",\n") . "        \"$dateStr\": $v";
-                $firstDate = false;
-            }
-            $buf .= "\n    }";
+            $buf .= $escapedPaths[$pid] . ": {\n" . \implode(",\n", $dateEntries) . "\n    }";
             \fwrite($file, $buf);
         }
 

From d5268d91c612331800328495d6920730e067ddd1 Mon Sep 17 00:00:00 2001
From: Lea Silakov <me@ileamare.ru>
Date: Fri, 13 Mar 2026 03:47:39 +0300
Subject: [PATCH 5/5] unwrapped the loop (14.3s)

---
 app/Parser.php | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/app/Parser.php b/app/Parser.php
index 36e79f72fa..5447179ed8 100644
--- a/app/Parser.php
+++ b/app/Parser.php
@@ -93,6 +93,42 @@ public function parse(string $inputPath, string $outputPath): void
             }
 
             $p = self::PREFIX_LEN;
+            $fence = $lastNl - 800;
+
+            while ($p < $fence) {
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+
+                $sep = \strpos($chunk, ',', $p);
+                $buckets[$pathIds[\substr($chunk, $p, $sep - $p)]] .= $dateIdBytes[\substr($chunk, $sep + 3, 8)];
+                $p = $sep + 52;
+            }
+
             while ($p < $lastNl) {
                 $sep = \strpos($chunk, ',', $p);
                 if ($sep === false || $sep >= $lastNl) break;